Building a classifier capable of determining a plant's species from a photo

import numpy as np   
import pandas as pd
import matplotlib.pyplot as plt 
%matplotlib inline 
import seaborn as sns

import os
import cv2
import warnings
import cv2 as cv
from glob import glob
from termcolor import colored
warnings.filterwarnings('ignore')

import itertools

import PIL.Image as Image

from tqdm import tqdm

import tensorflow as tf
import tensorflow_hub as hub
from tensorflow import keras
from keras.utils import np_utils
from tensorflow.keras import layers
from tensorflow.keras.optimizers import Adam
from tensorflow.keras.activations import softmax
from tensorflow.keras.utils import to_categorical
from tensorflow.keras import optimizers,regularizers
from tensorflow.python.keras.models import Sequential
from tensorflow.keras.layers import Reshape, Activation
from keras.callbacks import ModelCheckpoint, EarlyStopping
from keras.layers import Dropout,BatchNormalization, Activation
from tensorflow.keras.layers import Flatten, Dense, Conv2D, MaxPool2D
from tensorflow.python.keras.wrappers.scikit_learn import KerasRegressor



from sklearn import svm
from sklearn.svm import SVC
from sklearn import metrics
from sklearn.preprocessing import LabelEncoder
from sklearn.tree import DecisionTreeClassifier
from sklearn.ensemble import RandomForestClassifier
from sklearn.model_selection import train_test_split
from sklearn.metrics import confusion_matrix, classification_report

print(tf.__version__)

2.4.1

from google.colab import drive
drive.mount('/content/drive')

Mounted at /content/drive

print(colored('\x1B[1mImoprting dataset from drive.....','blue'))
pred1 = '/content/drive/MyDrive/Files/Part !/Part-1 - Plant Seedling Classification Data/Seedling - Prediction'
train = '/content/drive/MyDrive/Files/Part !/Part-1 - Plant Seedling Classification Data/train'

Imoprting dataset from drive.....

n_train = os.listdir(train)
n_pred = os.listdir(pred1)

print(colored('\x1B[1mGetting details of train dataset.....','blue'))
print('\x1B[1mTotal no of folders inside train dataset are',len(n_train))
print('\x1B[1mWhich are as follows:\n',n_train)

Getting details of train dataset.....
Total no of folders inside train dataset are 12
Which are as follows:
 ['Black-grass', 'Charlock', 'Cleavers', 'Common Chickweed', 'Common wheat', 'Fat Hen', 'Loose Silky-bent', 'Maize', 'Scentless Mayweed', 'Shepherds Purse', 'Small-flowered Cranesbill', 'Sugar beet']

print('\x1B[1mGetiing total no of imgaes from each folders','blue')
for folder in os.listdir(train):
  size=[]
  for image in os.listdir(train + '/' + folder):
    size.append(os.path.join(train, folder, image))
  print('\x1B[1mNo of images in',folder,':',len(size))

Getiing total no of imgaes from each folders blue
No of images in Black-grass : 263
No of images in Charlock : 390
No of images in Cleavers : 287
No of images in Common Chickweed : 611
No of images in Common wheat : 221
No of images in Fat Hen : 475
No of images in Loose Silky-bent : 671
No of images in Maize : 221
No of images in Scentless Mayweed : 516
No of images in Shepherds Purse : 231
No of images in Small-flowered Cranesbill : 496
No of images in Sugar beet : 385

classes = {}
for class_name in os.listdir(train):
    classes[class_name] = len(os.listdir(os.path.join(train, class_name)))

plt.bar(classes.keys(), classes.values())
plt.xticks(rotation=90)

([0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11],
 <a list of 12 Text major ticklabel objects>)

print(colored('\x1B[1mGetting random images from each folder of train dataset.....','blue'))

Getting random images from each folder of train dataset.....

import random
from PIL import Image 
from skimage.io import imread
imagee = [[] for x in range(1,13)]
label=[]
i=0
for folder in os.listdir(train):
  label.append(folder)
  for image in os.listdir(train + '/' + folder):
    imagee[i].append(os.path.join(train, folder, image))
  i=i+1
  
n = 0
i=0
for x in range(1,13):
  print(colored('\x1B[1mFolder Name','green'))
  print('\x1B[1m',label[i])
  plt.figure(figsize=(100, 100))
  plt.axis('off')
  for y in range(3):
    n += 1
    random_img = random.choice(imagee[i])
    imgs = imread(random_img)
    plt.subplot(18,18, n)
    plt.imshow(imgs)
  i += 1 
  plt.show()

Folder Name
 Black-grass

Folder Name
 Charlock

Folder Name
 Cleavers

Folder Name
 Common Chickweed

Folder Name
 Common wheat

Folder Name
 Fat Hen

Folder Name
 Loose Silky-bent

Folder Name
 Maize

Folder Name
 Scentless Mayweed

Folder Name
 Shepherds Purse

Folder Name
 Small-flowered Cranesbill

Folder Name
 Sugar beet

Seperating train Classes and images

images_path = '/content/drive/MyDrive/Files/Part !/Part-1 - Plant Seedling Classification Data/train/*/*.png'

print(colored('\x1B[1mSeperating Classes and Images of train dataset','blue'))
images = glob(images_path)
train_images = []
train_labels = []

for img in images:
    train_images.append(cv2.resize(cv2.imread(img), (64, 64)))
    train_labels.append(img.split('/')[-2])
train_X = np.asarray(train_images)
train_Y = pd.DataFrame(train_labels)

Seperating Classes and Images of train dataset

print(colored('\x1B[1mTotal numbers Images of train dataset','blue'))
print('\x1B[1m',len(train_X))

Total numbers Images of train dataset
 4767

print(colored('\x1B[1mShape of Images','blue'))
print('\x1B[1m',train_X.shape)

Shape of Images
 (4767, 64, 64, 3)

print(colored('\x1B[1mShape of Label','blue'))
print('\x1B[1m',train_Y.shape)

Shape of Label
 (4767, 1)

print(colored('\x1B[1mLabel encoding of Classes','blue'))

Label encoding of Classes

labels = LabelEncoder()
labels.fit(train_Y[0])
print('Classes'+str(labels.classes_))

Classes['Black-grass' 'Charlock' 'Cleavers' 'Common Chickweed' 'Common wheat'
 'Fat Hen' 'Loose Silky-bent' 'Maize' 'Scentless Mayweed'
 'Shepherds Purse' 'Small-flowered Cranesbill' 'Sugar beet']

y = labels.transform(train_Y[0])
classes = y.shape
print(str(classes))

(4767,)

#target_dict={k: v for v, k in enumerate(np.unique(train_Y[0]))}
#print(colored('\x1B[1mClasses  :  Labels','green'))
#target_dict

#target_val=  [target_dict[train_Y[0][i]] for i in range(len(train_Y[0]))]

#print(colored('\x1B[1mUnique labses','blue'))
#np.unique(target_val)

print(colored('\x1B[1mLength of total image set','blue'))
lenofimage = len(train_X)
lenofimage

Length of total image set

4767

print(colored('\x1B[1mReshaping of image dataset','blue'))
x = np.array(train_X).reshape(lenofimage,-1)
print('\x1B[1m',x.shape)

Reshaping of image dataset
 (4767, 12288)

print(colored('\x1B[1mMinimun and maximun Pixels of image','blue'))
print('Max' , x.max())
print('Min' , x.min())

Minimun and maximun Pixels of image
Max 255
Min 0

print(colored('\x1B[1mNormalzing','green'))
x = x/255 

Normalzing

print(colored('\x1B[1mMinimun and maximun Pixels of image','blue'))
print(colored('\x1B[1mAfter Normalzation','green'))
print('Max' , x.max())
print('Min' , x.min())

Minimun and maximun Pixels of image
After Normalzation
Max 1.0
Min 0.0

print(colored('\x1B[1mChanging target labels from list to array','blue'))
#y=np.array(target_val)
print('\x1B[1m',y.shape)

Changing target labels from list to array
 (4767,)

print(colored('\x1B[1mSpliting','blue'))
X_train, X_test, y_train, y_test = train_test_split(x, y, train_size=0.7, test_size=0.3, random_state=42)
score_train={}
score_test={}
accuracy={}
print(colored('\x1B[1mShape of datasets','green'))
print('Shape of X Train',X_train.shape)
print('Shape of Y Train',y_train.shape)
print('Shape of X Test',X_test.shape)
print('Shape of Y Test',y_test.shape)

Spliting
Shape of datasets
Shape of X Train (3336, 12288)
Shape of Y Train (3336,)
Shape of X Test (1431, 12288)
Shape of Y Test (1431,)

Model Building

I Supervised Learning

1. SVM

2. Random Forest

3. Decision Tree

II Neural Network

III CNN

I Supervised Learning

1. SVM

print(colored('\x1B[1mSVM','blue'))
svm = SVC(kernel='linear', gamma= 'auto',C=.1)

SVM

print(colored('\x1B[1mFitting Model','blue'))
svm.fit(X_train, y_train)

Fitting Model

SVC(C=0.1, break_ties=False, cache_size=200, class_weight=None, coef0=0.0,
    decision_function_shape='ovr', degree=3, gamma='auto', kernel='linear',
    max_iter=-1, probability=False, random_state=None, shrinking=True,
    tol=0.001, verbose=False)

print(colored('\x1B[1mPredicition','blue'))
y_pred = svm.predict(X_test)

model_train = svm.score(X_train,y_train)
model_test = svm.score(X_test,y_test)
score_train['SVM'] = model_train
score_test['SVM Test Score']=model_test
accuracy['SVM Accuracy'] = metrics.accuracy_score(y_test, y_pred)

Predicition

print(colored('\x1B[1mAccuracy','blue'))
print('Train Score:',model_train)
print('Test Score:',model_test)

Accuracy
Train Score: 0.9961031175059952
Test Score: 0.4996505939902166

print(colored('\x1B[1mClassification Report','blue'))
print(classification_report(y_test, y_pred))
cm_svc = (y_test != y_pred).sum()
print('\n Misclassified samples in SVM: {}'.format(cm_svc))

Classification Report
              precision    recall  f1-score   support

           0       0.32      0.24      0.27        89
           1       0.60      0.64      0.62       120
           2       0.58      0.57      0.57        86
           3       0.54      0.64      0.59       190
           4       0.17      0.20      0.18        59
           5       0.41      0.47      0.44       137
           6       0.50      0.64      0.56       196
           7       0.69      0.28      0.40        72
           8       0.45      0.52      0.48       143
           9       0.38      0.32      0.35        62
          10       0.70      0.66      0.68       154
          11       0.49      0.24      0.33       123

    accuracy                           0.50      1431
   macro avg       0.49      0.45      0.46      1431
weighted avg       0.51      0.50      0.49      1431


 Misclassified samples in SVM: 716

I Supervised Learning

2. Random Forest

print(colored('\x1B[1mRandom Forest','blue'))
rf = RandomForestClassifier(n_estimators= 25, max_features= 6, max_depth= 10, bootstrap= False)

Random Forest

print(colored('\x1B[1mFitting Model','blue'))
rf.fit(X_train,y_train)

Fitting Model

RandomForestClassifier(bootstrap=False, ccp_alpha=0.0, class_weight=None,
                       criterion='gini', max_depth=10, max_features=6,
                       max_leaf_nodes=None, max_samples=None,
                       min_impurity_decrease=0.0, min_impurity_split=None,
                       min_samples_leaf=1, min_samples_split=2,
                       min_weight_fraction_leaf=0.0, n_estimators=25,
                       n_jobs=None, oob_score=False, random_state=None,
                       verbose=0, warm_start=False)

print(colored('\x1B[1mPredicition','blue'))
y_pred = rf.predict(X_test)
model_train = rf.score(X_train,y_train)
model_test = rf.score(X_test,y_test)
score_train['Random Forest'] = model_train
score_test['Random Forest Test Score']=model_test
accuracy['Random Forest Accuracy'] = metrics.accuracy_score(y_test, y_pred)

Predicition

print(colored('\x1B[1mAccuracy','blue'))
print('Train Score:',model_train)
print('Test Score:',model_test)

Accuracy
Train Score: 0.9589328537170264
Test Score: 0.4276729559748428

print(colored('\x1B[1mClassification Report','blue'))
print(classification_report(y_test, y_pred))
cm_rf = (y_test != y_pred).sum()
print('\n Misclassified samples in Random Forest: {}'.format(cm_rf))

Classification Report
              precision    recall  f1-score   support

           0       0.67      0.04      0.08        89
           1       0.56      0.40      0.47       120
           2       0.58      0.29      0.39        86
           3       0.51      0.76      0.61       190
           4       0.33      0.02      0.03        59
           5       0.34      0.20      0.26       137
           6       0.33      0.87      0.48       196
           7       0.82      0.12      0.22        72
           8       0.42      0.56      0.48       143
           9       0.00      0.00      0.00        62
          10       0.50      0.61      0.55       154
          11       0.50      0.07      0.12       123

    accuracy                           0.43      1431
   macro avg       0.46      0.33      0.31      1431
weighted avg       0.46      0.43      0.37      1431


 Misclassified samples in Random Forest: 819

I Supervised Learning

3. Decision Tree

print(colored('\x1B[1mCDecision Tree','blue'))
dTree = DecisionTreeClassifier(min_samples_leaf=7, max_features=5, max_depth=5, criterion='gini', random_state=1)

CDecision Tree

print(colored('\x1B[1mFitting Model','blue'))
dTree.fit(X_train,y_train)

Fitting Model

DecisionTreeClassifier(ccp_alpha=0.0, class_weight=None, criterion='gini',
                       max_depth=5, max_features=5, max_leaf_nodes=None,
                       min_impurity_decrease=0.0, min_impurity_split=None,
                       min_samples_leaf=7, min_samples_split=2,
                       min_weight_fraction_leaf=0.0, presort='deprecated',
                       random_state=1, splitter='best')

print(colored('\x1B[1mPredicition','blue'))
y_pred = dTree.predict(X_test)
model_train = dTree.score(X_train,y_train)
model_test = dTree.score(X_test,y_test)
score_train['Decision Tree'] = model_train
score_test['Decision Tree Test Score']=model_test
accuracy['Decision Tree Accuracy'] = metrics.accuracy_score(y_test, y_pred)

Predicition

print(colored('\x1B[1mAccuracy','blue'))
print('Train Score:',model_train)
print('Test Score:',model_test)

Accuracy
Train Score: 0.2643884892086331
Test Score: 0.2236198462613557

print(colored('\x1B[1mClassification Report','blue'))
print(classification_report(y_test, y_pred))
cm_dTree = (y_test != y_pred).sum()
print('\n Misclassified samples in Decision Tree: {}'.format(cm_dTree))

Classification Report
              precision    recall  f1-score   support

           0       0.00      0.00      0.00        89
           1       0.20      0.02      0.03       120
           2       0.00      0.00      0.00        86
           3       0.37      0.32      0.34       190
           4       0.00      0.00      0.00        59
           5       0.10      0.04      0.06       137
           6       0.20      0.74      0.31       196
           7       0.26      0.10      0.14        72
           8       0.18      0.34      0.23       143
           9       0.00      0.00      0.00        62
          10       0.31      0.32      0.32       154
          11       0.00      0.00      0.00       123

    accuracy                           0.22      1431
   macro avg       0.13      0.16      0.12      1431
weighted avg       0.17      0.22      0.16      1431


 Misclassified samples in Decision Tree: 1111

II Neural Network

print(colored('\x1B[1mSetting labels to categorically labels','blue'))
y_c = to_categorical(y, num_classes=12)

Setting labels to categorically labels

print(colored('\x1B[1mSpliting','blue'))
X_train, X_test, y_train, y_test = train_test_split(x, y_c, train_size=0.8, test_size=0.2, random_state=42)

Spliting

print(colored('\x1B[1mShape of datasets','green'))
print('Shape of X Train',X_train.shape)
print('Shape of Y Train',y_train.shape)
print('Shape of X Test',X_test.shape)
print('Shape of Y Test',y_test.shape)

Shape of datasets
Shape of X Train (3813, 12288)
Shape of Y Train (3813, 12)
Shape of X Test (954, 12288)
Shape of Y Test (954, 12)

input = x.shape[1]
input

12288

print(colored('\x1B[1mNeural Network Model Building','blue'))

Neural Network Model Building

hidden_nodes = 256
output_nodes = 12

nn_sgd = Sequential()
nn_sgd.add(Dense(hidden_nodes, input_shape=(input,), activation='relu'))
nn_sgd.add(Dense(hidden_nodes, activation='relu'))
nn_sgd.add(Dense(output_nodes, activation='softmax', kernel_regularizer=regularizers.l2(0)))

print(colored('\x1B[1mCompiling Model Using SGD Optimizer','blue'))
sgd = optimizers.SGD(lr=.001, decay=1e-6, momentum=0.9)
nn_sgd.compile(loss='categorical_crossentropy', optimizer=sgd, metrics=['accuracy'])

Compiling Model Using SGD Optimizer

print(colored('\x1B[1mFitting Model Using SGD Optimizer','blue'))

Fitting Model Using SGD Optimizer

nn_sgd.fit(X_train, y_train, epochs=100, batch_size=200, verbose= 1)
score = nn_sgd.evaluate(X_test, y_test, verbose=0)
score_train['NN(SGD)'] = nn_sgd.evaluate(X_train, y_train, verbose=0)
score_test['NN(SGD) Test Score']=score
accuracy['NN(SGD) Accuracy'] = score

Epoch 1/100
20/20 [==============================] - 2s 69ms/step - loss: 2.4953 - accuracy: 0.1348
Epoch 2/100
20/20 [==============================] - 1s 75ms/step - loss: 2.4016 - accuracy: 0.1637
Epoch 3/100
20/20 [==============================] - 2s 85ms/step - loss: 2.3854 - accuracy: 0.1758
Epoch 4/100
20/20 [==============================] - 2s 77ms/step - loss: 2.3631 - accuracy: 0.1844
Epoch 5/100
20/20 [==============================] - 2s 79ms/step - loss: 2.3549 - accuracy: 0.1942
Epoch 6/100
20/20 [==============================] - 2s 91ms/step - loss: 2.3278 - accuracy: 0.2308
Epoch 7/100
20/20 [==============================] - 2s 94ms/step - loss: 2.3116 - accuracy: 0.2218
Epoch 8/100
20/20 [==============================] - 2s 89ms/step - loss: 2.2991 - accuracy: 0.2487
Epoch 9/100
20/20 [==============================] - 2s 97ms/step - loss: 2.2730 - accuracy: 0.2410
Epoch 10/100
20/20 [==============================] - 2s 80ms/step - loss: 2.2665 - accuracy: 0.2646
Epoch 11/100
20/20 [==============================] - 2s 88ms/step - loss: 2.2390 - accuracy: 0.2572
Epoch 12/100
20/20 [==============================] - 2s 74ms/step - loss: 2.2312 - accuracy: 0.2588
Epoch 13/100
20/20 [==============================] - 1s 66ms/step - loss: 2.1925 - accuracy: 0.2934
Epoch 14/100
20/20 [==============================] - 1s 67ms/step - loss: 2.1931 - accuracy: 0.2531
Epoch 15/100
20/20 [==============================] - 1s 66ms/step - loss: 2.1741 - accuracy: 0.3064
Epoch 16/100
20/20 [==============================] - 1s 67ms/step - loss: 2.1579 - accuracy: 0.3089
Epoch 17/100
20/20 [==============================] - 1s 67ms/step - loss: 2.1377 - accuracy: 0.2940
Epoch 18/100
20/20 [==============================] - 1s 67ms/step - loss: 2.1174 - accuracy: 0.2964
Epoch 19/100
20/20 [==============================] - 1s 68ms/step - loss: 2.0909 - accuracy: 0.3009
Epoch 20/100
20/20 [==============================] - 1s 67ms/step - loss: 2.0744 - accuracy: 0.3084
Epoch 21/100
20/20 [==============================] - 1s 69ms/step - loss: 2.0502 - accuracy: 0.3556
Epoch 22/100
20/20 [==============================] - 1s 67ms/step - loss: 2.0338 - accuracy: 0.3534
Epoch 23/100
20/20 [==============================] - 1s 66ms/step - loss: 2.0060 - accuracy: 0.3352
Epoch 24/100
20/20 [==============================] - 1s 68ms/step - loss: 1.9850 - accuracy: 0.3545
Epoch 25/100
20/20 [==============================] - 1s 66ms/step - loss: 1.9625 - accuracy: 0.3634
Epoch 26/100
20/20 [==============================] - 1s 66ms/step - loss: 1.9484 - accuracy: 0.3575
Epoch 27/100
20/20 [==============================] - 1s 67ms/step - loss: 1.9206 - accuracy: 0.3757
Epoch 28/100
20/20 [==============================] - 1s 69ms/step - loss: 1.9234 - accuracy: 0.3442
Epoch 29/100
20/20 [==============================] - 1s 67ms/step - loss: 1.8855 - accuracy: 0.3930
Epoch 30/100
20/20 [==============================] - 1s 68ms/step - loss: 1.8668 - accuracy: 0.3888
Epoch 31/100
20/20 [==============================] - 1s 68ms/step - loss: 1.8501 - accuracy: 0.3671
Epoch 32/100
20/20 [==============================] - 1s 67ms/step - loss: 1.8188 - accuracy: 0.4092
Epoch 33/100
20/20 [==============================] - 1s 67ms/step - loss: 1.8112 - accuracy: 0.3784
Epoch 34/100
20/20 [==============================] - 1s 68ms/step - loss: 1.7794 - accuracy: 0.4215
Epoch 35/100
20/20 [==============================] - 1s 67ms/step - loss: 1.7540 - accuracy: 0.4222
Epoch 36/100
20/20 [==============================] - 1s 66ms/step - loss: 1.7495 - accuracy: 0.4318
Epoch 37/100
20/20 [==============================] - 1s 67ms/step - loss: 1.7194 - accuracy: 0.4556
Epoch 38/100
20/20 [==============================] - 1s 66ms/step - loss: 1.7080 - accuracy: 0.4643
Epoch 39/100
20/20 [==============================] - 1s 68ms/step - loss: 1.7043 - accuracy: 0.4450
Epoch 40/100
20/20 [==============================] - 1s 66ms/step - loss: 1.6639 - accuracy: 0.4798
Epoch 41/100
20/20 [==============================] - 1s 68ms/step - loss: 1.6470 - accuracy: 0.4607
Epoch 42/100
20/20 [==============================] - 1s 67ms/step - loss: 1.6292 - accuracy: 0.4864
Epoch 43/100
20/20 [==============================] - 1s 69ms/step - loss: 1.6036 - accuracy: 0.4697
Epoch 44/100
20/20 [==============================] - 1s 68ms/step - loss: 1.5782 - accuracy: 0.4781
Epoch 45/100
20/20 [==============================] - 1s 67ms/step - loss: 1.5685 - accuracy: 0.5002
Epoch 46/100
20/20 [==============================] - 1s 67ms/step - loss: 1.5803 - accuracy: 0.4808
Epoch 47/100
20/20 [==============================] - 1s 66ms/step - loss: 1.5608 - accuracy: 0.4856
Epoch 48/100
20/20 [==============================] - 1s 66ms/step - loss: 1.5293 - accuracy: 0.5034
Epoch 49/100
20/20 [==============================] - 1s 67ms/step - loss: 1.5230 - accuracy: 0.5133
Epoch 50/100
20/20 [==============================] - 1s 67ms/step - loss: 1.4886 - accuracy: 0.5204
Epoch 51/100
20/20 [==============================] - 1s 68ms/step - loss: 1.4853 - accuracy: 0.5318
Epoch 52/100
20/20 [==============================] - 1s 67ms/step - loss: 1.4585 - accuracy: 0.5456
Epoch 53/100
20/20 [==============================] - 1s 66ms/step - loss: 1.4825 - accuracy: 0.5193
Epoch 54/100
20/20 [==============================] - 1s 68ms/step - loss: 1.4263 - accuracy: 0.5577
Epoch 55/100
20/20 [==============================] - 1s 67ms/step - loss: 1.4306 - accuracy: 0.5306
Epoch 56/100
20/20 [==============================] - 1s 68ms/step - loss: 1.3999 - accuracy: 0.5628
Epoch 57/100
20/20 [==============================] - 1s 66ms/step - loss: 1.4110 - accuracy: 0.5601
Epoch 58/100
20/20 [==============================] - 1s 68ms/step - loss: 1.3995 - accuracy: 0.5525
Epoch 59/100
20/20 [==============================] - 1s 69ms/step - loss: 1.3906 - accuracy: 0.5358
Epoch 60/100
20/20 [==============================] - 1s 66ms/step - loss: 1.3502 - accuracy: 0.5680
Epoch 61/100
20/20 [==============================] - 1s 68ms/step - loss: 1.3103 - accuracy: 0.5887
Epoch 62/100
20/20 [==============================] - 1s 67ms/step - loss: 1.3336 - accuracy: 0.5634
Epoch 63/100
20/20 [==============================] - 1s 69ms/step - loss: 1.3177 - accuracy: 0.5620
Epoch 64/100
20/20 [==============================] - 1s 67ms/step - loss: 1.3360 - accuracy: 0.5557
Epoch 65/100
20/20 [==============================] - 1s 68ms/step - loss: 1.2880 - accuracy: 0.5897
Epoch 66/100
20/20 [==============================] - 1s 68ms/step - loss: 1.3079 - accuracy: 0.5626
Epoch 67/100
20/20 [==============================] - 1s 68ms/step - loss: 1.2862 - accuracy: 0.5927
Epoch 68/100
20/20 [==============================] - 1s 66ms/step - loss: 1.2471 - accuracy: 0.5988
Epoch 69/100
20/20 [==============================] - 1s 67ms/step - loss: 1.2611 - accuracy: 0.6022
Epoch 70/100
20/20 [==============================] - 1s 66ms/step - loss: 1.2345 - accuracy: 0.6059
Epoch 71/100
20/20 [==============================] - 1s 67ms/step - loss: 1.2414 - accuracy: 0.6153
Epoch 72/100
20/20 [==============================] - 1s 69ms/step - loss: 1.2243 - accuracy: 0.6130
Epoch 73/100
20/20 [==============================] - 1s 67ms/step - loss: 1.2039 - accuracy: 0.6033
Epoch 74/100
20/20 [==============================] - 1s 69ms/step - loss: 1.1976 - accuracy: 0.6258
Epoch 75/100
20/20 [==============================] - 1s 68ms/step - loss: 1.1859 - accuracy: 0.6187
Epoch 76/100
20/20 [==============================] - 1s 67ms/step - loss: 1.1939 - accuracy: 0.6155
Epoch 77/100
20/20 [==============================] - 1s 66ms/step - loss: 1.1608 - accuracy: 0.6454
Epoch 78/100
20/20 [==============================] - 1s 67ms/step - loss: 1.1507 - accuracy: 0.6409
Epoch 79/100
20/20 [==============================] - 1s 67ms/step - loss: 1.1386 - accuracy: 0.6440
Epoch 80/100
20/20 [==============================] - 1s 66ms/step - loss: 1.1183 - accuracy: 0.6576
Epoch 81/100
20/20 [==============================] - 1s 67ms/step - loss: 1.1076 - accuracy: 0.6658
Epoch 82/100
20/20 [==============================] - 1s 67ms/step - loss: 1.1094 - accuracy: 0.6642
Epoch 83/100
20/20 [==============================] - 1s 67ms/step - loss: 1.1435 - accuracy: 0.6246
Epoch 84/100
20/20 [==============================] - 1s 67ms/step - loss: 1.0996 - accuracy: 0.6551
Epoch 85/100
20/20 [==============================] - 1s 67ms/step - loss: 1.1018 - accuracy: 0.6417
Epoch 86/100
20/20 [==============================] - 1s 68ms/step - loss: 1.0779 - accuracy: 0.6615
Epoch 87/100
20/20 [==============================] - 1s 69ms/step - loss: 1.1202 - accuracy: 0.6169
Epoch 88/100
20/20 [==============================] - 1s 68ms/step - loss: 1.0727 - accuracy: 0.6720
Epoch 89/100
20/20 [==============================] - 1s 67ms/step - loss: 1.0874 - accuracy: 0.6334
Epoch 90/100
20/20 [==============================] - 1s 67ms/step - loss: 1.0548 - accuracy: 0.6738
Epoch 91/100
20/20 [==============================] - 1s 66ms/step - loss: 1.0263 - accuracy: 0.6743
Epoch 92/100
20/20 [==============================] - 1s 67ms/step - loss: 1.0284 - accuracy: 0.6688
Epoch 93/100
20/20 [==============================] - 1s 67ms/step - loss: 0.9980 - accuracy: 0.6903
Epoch 94/100
20/20 [==============================] - 1s 66ms/step - loss: 0.9948 - accuracy: 0.6909
Epoch 95/100
20/20 [==============================] - 1s 68ms/step - loss: 0.9704 - accuracy: 0.7062
Epoch 96/100
20/20 [==============================] - 1s 67ms/step - loss: 1.0265 - accuracy: 0.6673
Epoch 97/100
20/20 [==============================] - 1s 68ms/step - loss: 1.0043 - accuracy: 0.6783
Epoch 98/100
20/20 [==============================] - 1s 67ms/step - loss: 0.9969 - accuracy: 0.6876
Epoch 99/100
20/20 [==============================] - 1s 67ms/step - loss: 0.9652 - accuracy: 0.6864
Epoch 100/100
20/20 [==============================] - 1s 67ms/step - loss: 0.9836 - accuracy: 0.6717

print(colored('\x1B[1mScore','green'),score)

Score [1.4598736763000488, 0.5073375105857849]

hidden_nodes = 256
output_nodes = 12

nn_adam = Sequential()
nn_adam.add(Dense(hidden_nodes, input_shape=(input,), activation='relu'))
nn_adam.add(Dense(hidden_nodes, activation='relu'))
nn_adam.add(Dense(output_nodes, activation='softmax', kernel_regularizer=regularizers.l2(0)))

print(colored('\x1B[1mCompiling Model Using ADAM Optimizer','blue'))
opt = Adam(lr=0.001, beta_1=0.9, beta_2=0.999, epsilon=None, decay=0.001, amsgrad=False)
nn_adam.compile(optimizer = opt, loss = 'categorical_crossentropy', metrics = ['accuracy'])

Compiling Model Using ADAM Optimizer

print(colored('\x1B[1mFitting Model Using ADAM Optimizer','blue'))

Fitting Model Using ADAM Optimizer

nn_adam.fit(X_train, y_train, epochs=100, batch_size=200, verbose= 1)
score = nn_adam.evaluate(X_test, y_test, verbose=0)
score_train['NN(ADAM)'] = nn_adam.evaluate(X_train, y_train, verbose=0)
score_test['NN Test Score']=score
accuracy['NN Accuracy'] = score

Epoch 1/100
20/20 [==============================] - 2s 71ms/step - loss: 3.6157 - accuracy: 0.1136
Epoch 2/100
20/20 [==============================] - 1s 72ms/step - loss: 2.4529 - accuracy: 0.1326
Epoch 3/100
20/20 [==============================] - 1s 71ms/step - loss: 2.3846 - accuracy: 0.1802
Epoch 4/100
20/20 [==============================] - 1s 70ms/step - loss: 2.3551 - accuracy: 0.1613
Epoch 5/100
20/20 [==============================] - 1s 71ms/step - loss: 2.2313 - accuracy: 0.2386
Epoch 6/100
20/20 [==============================] - 1s 70ms/step - loss: 2.1137 - accuracy: 0.2815
Epoch 7/100
20/20 [==============================] - 1s 70ms/step - loss: 2.0997 - accuracy: 0.2442
Epoch 8/100
20/20 [==============================] - 1s 71ms/step - loss: 1.9709 - accuracy: 0.3301
Epoch 9/100
20/20 [==============================] - 1s 69ms/step - loss: 1.9637 - accuracy: 0.3265
Epoch 10/100
20/20 [==============================] - 1s 69ms/step - loss: 1.8187 - accuracy: 0.3911
Epoch 11/100
20/20 [==============================] - 1s 70ms/step - loss: 1.7756 - accuracy: 0.3989
Epoch 12/100
20/20 [==============================] - 1s 68ms/step - loss: 1.7380 - accuracy: 0.3934
Epoch 13/100
20/20 [==============================] - 1s 69ms/step - loss: 1.6948 - accuracy: 0.4015
Epoch 14/100
20/20 [==============================] - 1s 70ms/step - loss: 1.6987 - accuracy: 0.3790
Epoch 15/100
20/20 [==============================] - 1s 70ms/step - loss: 1.5841 - accuracy: 0.4599
Epoch 16/100
20/20 [==============================] - 1s 69ms/step - loss: 1.4903 - accuracy: 0.4963
Epoch 17/100
20/20 [==============================] - 1s 71ms/step - loss: 1.6217 - accuracy: 0.4249
Epoch 18/100
20/20 [==============================] - 1s 71ms/step - loss: 1.4456 - accuracy: 0.5085
Epoch 19/100
20/20 [==============================] - 1s 70ms/step - loss: 1.4961 - accuracy: 0.4865
Epoch 20/100
20/20 [==============================] - 1s 69ms/step - loss: 1.3324 - accuracy: 0.5608
Epoch 21/100
20/20 [==============================] - 1s 68ms/step - loss: 1.2951 - accuracy: 0.5615
Epoch 22/100
20/20 [==============================] - 1s 69ms/step - loss: 1.3330 - accuracy: 0.5269
Epoch 23/100
20/20 [==============================] - 1s 71ms/step - loss: 1.2664 - accuracy: 0.5614
Epoch 24/100
20/20 [==============================] - 1s 69ms/step - loss: 1.2051 - accuracy: 0.6016
Epoch 25/100
20/20 [==============================] - 1s 69ms/step - loss: 1.2675 - accuracy: 0.5499
Epoch 26/100
20/20 [==============================] - 1s 70ms/step - loss: 1.2458 - accuracy: 0.5546
Epoch 27/100
20/20 [==============================] - 1s 68ms/step - loss: 1.2803 - accuracy: 0.5324
Epoch 28/100
20/20 [==============================] - 1s 69ms/step - loss: 1.2054 - accuracy: 0.5788
Epoch 29/100
20/20 [==============================] - 1s 70ms/step - loss: 1.0915 - accuracy: 0.6314
Epoch 30/100
20/20 [==============================] - 1s 71ms/step - loss: 1.0891 - accuracy: 0.6306
Epoch 31/100
20/20 [==============================] - 1s 70ms/step - loss: 1.0302 - accuracy: 0.6514
Epoch 32/100
20/20 [==============================] - 1s 69ms/step - loss: 0.9949 - accuracy: 0.6721
Epoch 33/100
20/20 [==============================] - 1s 70ms/step - loss: 1.0147 - accuracy: 0.6534
Epoch 34/100
20/20 [==============================] - 1s 70ms/step - loss: 0.9788 - accuracy: 0.6743
Epoch 35/100
20/20 [==============================] - 1s 68ms/step - loss: 0.9478 - accuracy: 0.6915
Epoch 36/100
20/20 [==============================] - 1s 69ms/step - loss: 0.9492 - accuracy: 0.6786
Epoch 37/100
20/20 [==============================] - 1s 71ms/step - loss: 1.0141 - accuracy: 0.6357
Epoch 38/100
20/20 [==============================] - 1s 70ms/step - loss: 0.9619 - accuracy: 0.6648
Epoch 39/100
20/20 [==============================] - 1s 68ms/step - loss: 0.9828 - accuracy: 0.6615
Epoch 40/100
20/20 [==============================] - 1s 70ms/step - loss: 0.8419 - accuracy: 0.7332
Epoch 41/100
20/20 [==============================] - 1s 68ms/step - loss: 0.8877 - accuracy: 0.6989
Epoch 42/100
20/20 [==============================] - 1s 69ms/step - loss: 0.7982 - accuracy: 0.7482
Epoch 43/100
20/20 [==============================] - 1s 71ms/step - loss: 0.8494 - accuracy: 0.7166
Epoch 44/100
20/20 [==============================] - 1s 69ms/step - loss: 0.8278 - accuracy: 0.7361
Epoch 45/100
20/20 [==============================] - 1s 71ms/step - loss: 0.7907 - accuracy: 0.7431
Epoch 46/100
20/20 [==============================] - 1s 69ms/step - loss: 0.8065 - accuracy: 0.7263
Epoch 47/100
20/20 [==============================] - 1s 69ms/step - loss: 0.8037 - accuracy: 0.7439
Epoch 48/100
20/20 [==============================] - 1s 69ms/step - loss: 0.7599 - accuracy: 0.7542
Epoch 49/100
20/20 [==============================] - 1s 71ms/step - loss: 0.7791 - accuracy: 0.7432
Epoch 50/100
20/20 [==============================] - 1s 69ms/step - loss: 0.7711 - accuracy: 0.7474
Epoch 51/100
20/20 [==============================] - 1s 70ms/step - loss: 0.7134 - accuracy: 0.7716
Epoch 52/100
20/20 [==============================] - 1s 69ms/step - loss: 0.7395 - accuracy: 0.7509
Epoch 53/100
20/20 [==============================] - 1s 70ms/step - loss: 0.7052 - accuracy: 0.7806
Epoch 54/100
20/20 [==============================] - 1s 70ms/step - loss: 0.6933 - accuracy: 0.7769
Epoch 55/100
20/20 [==============================] - 1s 69ms/step - loss: 0.7746 - accuracy: 0.7246
Epoch 56/100
20/20 [==============================] - 1s 70ms/step - loss: 0.6652 - accuracy: 0.7892
Epoch 57/100
20/20 [==============================] - 1s 70ms/step - loss: 0.6600 - accuracy: 0.7943
Epoch 58/100
20/20 [==============================] - 1s 72ms/step - loss: 0.6216 - accuracy: 0.8175
Epoch 59/100
20/20 [==============================] - 1s 71ms/step - loss: 0.6093 - accuracy: 0.8134
Epoch 60/100
20/20 [==============================] - 1s 69ms/step - loss: 0.6283 - accuracy: 0.8010
Epoch 61/100
20/20 [==============================] - 1s 70ms/step - loss: 0.6310 - accuracy: 0.8016
Epoch 62/100
20/20 [==============================] - 1s 70ms/step - loss: 0.6031 - accuracy: 0.8096
Epoch 63/100
20/20 [==============================] - 1s 69ms/step - loss: 0.5697 - accuracy: 0.8371
Epoch 64/100
20/20 [==============================] - 1s 71ms/step - loss: 0.6653 - accuracy: 0.7797
Epoch 65/100
20/20 [==============================] - 1s 69ms/step - loss: 0.5718 - accuracy: 0.8313
Epoch 66/100
20/20 [==============================] - 1s 71ms/step - loss: 0.6031 - accuracy: 0.8164
Epoch 67/100
20/20 [==============================] - 1s 70ms/step - loss: 0.6216 - accuracy: 0.7900
Epoch 68/100
20/20 [==============================] - 1s 69ms/step - loss: 0.5554 - accuracy: 0.8277
Epoch 69/100
20/20 [==============================] - 1s 73ms/step - loss: 0.5220 - accuracy: 0.8408
Epoch 70/100
20/20 [==============================] - 1s 70ms/step - loss: 0.5010 - accuracy: 0.8524
Epoch 71/100
20/20 [==============================] - 1s 71ms/step - loss: 0.5524 - accuracy: 0.8252
Epoch 72/100
20/20 [==============================] - 1s 70ms/step - loss: 0.5036 - accuracy: 0.8505
Epoch 73/100
20/20 [==============================] - 1s 71ms/step - loss: 0.5122 - accuracy: 0.8370
Epoch 74/100
20/20 [==============================] - 1s 71ms/step - loss: 0.4898 - accuracy: 0.8495
Epoch 75/100
20/20 [==============================] - 1s 71ms/step - loss: 0.4524 - accuracy: 0.8786
Epoch 76/100
20/20 [==============================] - 1s 70ms/step - loss: 0.4848 - accuracy: 0.8617
Epoch 77/100
20/20 [==============================] - 1s 70ms/step - loss: 0.4798 - accuracy: 0.8611
Epoch 78/100
20/20 [==============================] - 1s 69ms/step - loss: 0.4656 - accuracy: 0.8654
Epoch 79/100
20/20 [==============================] - 1s 72ms/step - loss: 0.4926 - accuracy: 0.8481
Epoch 80/100
20/20 [==============================] - 1s 73ms/step - loss: 0.5299 - accuracy: 0.8291
Epoch 81/100
20/20 [==============================] - 1s 71ms/step - loss: 0.4649 - accuracy: 0.8775
Epoch 82/100
20/20 [==============================] - 1s 71ms/step - loss: 0.4097 - accuracy: 0.8877
Epoch 83/100
20/20 [==============================] - 1s 70ms/step - loss: 0.4409 - accuracy: 0.8707
Epoch 84/100
20/20 [==============================] - 1s 72ms/step - loss: 0.4433 - accuracy: 0.8719
Epoch 85/100
20/20 [==============================] - 1s 70ms/step - loss: 0.4039 - accuracy: 0.8942
Epoch 86/100
20/20 [==============================] - 1s 72ms/step - loss: 0.4237 - accuracy: 0.8753
Epoch 87/100
20/20 [==============================] - 1s 70ms/step - loss: 0.4851 - accuracy: 0.8494
Epoch 88/100
20/20 [==============================] - 1s 69ms/step - loss: 0.5060 - accuracy: 0.8289
Epoch 89/100
20/20 [==============================] - 1s 71ms/step - loss: 0.4024 - accuracy: 0.8805
Epoch 90/100
20/20 [==============================] - 1s 70ms/step - loss: 0.3918 - accuracy: 0.8828
Epoch 91/100
20/20 [==============================] - 1s 72ms/step - loss: 0.3782 - accuracy: 0.9016
Epoch 92/100
20/20 [==============================] - 1s 72ms/step - loss: 0.3629 - accuracy: 0.9091
Epoch 93/100
20/20 [==============================] - 1s 71ms/step - loss: 0.3524 - accuracy: 0.9146
Epoch 94/100
20/20 [==============================] - 1s 73ms/step - loss: 0.3468 - accuracy: 0.9145
Epoch 95/100
20/20 [==============================] - 1s 72ms/step - loss: 0.3539 - accuracy: 0.9099
Epoch 96/100
20/20 [==============================] - 1s 73ms/step - loss: 0.3382 - accuracy: 0.9169
Epoch 97/100
20/20 [==============================] - 1s 70ms/step - loss: 0.3594 - accuracy: 0.9004
Epoch 98/100
20/20 [==============================] - 1s 71ms/step - loss: 0.3392 - accuracy: 0.9122
Epoch 99/100
20/20 [==============================] - 1s 69ms/step - loss: 0.3170 - accuracy: 0.9285
Epoch 100/100
20/20 [==============================] - 1s 69ms/step - loss: 0.3363 - accuracy: 0.9182

print(colored('\x1B[1mScore','green'),score)

Score [1.6843140125274658, 0.5094339847564697]

III CNN

print(colored('\x1B[1mSpliting Train and Test sets','blue'))
X_train,X_test,y_train,y_test=train_test_split(train_X,y_c,test_size=0.2,random_state=7)

Spliting Train and Test sets

print(colored('\x1B[1mShape of datasets','green'))
print('Shape of X Train',X_train.shape)
print('Shape of Y Train',y_train.shape)

Shape of datasets
Shape of X Train (3813, 64, 64, 3)
Shape of Y Train (3813, 12)

print(colored('\x1B[1mSpliting Test and Validation sets','blue'))
X_val,X_test_new,y_val,y_test_new = train_test_split(X_test,y_test,test_size=0.2,random_state=38,stratify=y_test)

Spliting Test and Validation sets

print(colored('\x1B[1mShape of datasets','green'))
print('Shape of X_val',X_val.shape)
print('Shape of Y_val',y_val.shape)
print('Shape of X Test',X_test_new.shape)
print('Shape of Y Test',y_test_new.shape)

Shape of datasets
Shape of X_val (763, 64, 64, 3)
Shape of Y_val (763, 12)
Shape of X Test (191, 64, 64, 3)
Shape of Y Test (191, 12)

print(colored('\x1B[1mPreprocessing Image and applying augmentation','blue'))

image_generator_train = tf.keras.preprocessing.image.ImageDataGenerator(
        rotation_range=20,width_shift_range=0.2,height_shift_range=0.2,
        shear_range=0.3,zoom_range=0.5,horizontal_flip=True,vertical_flip=True,
        brightness_range=[0.2,0.8],featurewise_center=False,samplewise_center=False,
        featurewise_std_normalization=False,samplewise_std_normalization=False,
        fill_mode = "reflect")
image_generator_train.fit(X_train, augment=True)
augmented_train = image_generator_train.flow(np.array(X_train),np.array(y_train))

image_generator_val = tf.keras.preprocessing.image.ImageDataGenerator(
        rotation_range=20,width_shift_range=0.2,height_shift_range=0.2,
        shear_range=0.3,zoom_range=0.5,horizontal_flip=True,vertical_flip=True,
        brightness_range=[0.2,0.8],featurewise_center=False,samplewise_center=False,
        featurewise_std_normalization=False,samplewise_std_normalization=False,
        fill_mode = "reflect")
image_generator_val.fit(X_val, augment=True)
augmented_val = image_generator_val.flow(np.array(X_val),np.array(y_val))

image_generator_test = tf.keras.preprocessing.image.ImageDataGenerator(rescale = 1./255)
image_generator_test.fit(X_test)
augmented_test = image_generator_test.flow(np.array(X_test),np.array(y_test))

print(colored('\x1B[1mFitting Train set in Preprocessing and Augmentation featuers','green'))

Preprocessing Image and applying augmentation
Fitting Train set in Preprocessing and Augmentation featuers

print(colored('\x1B[1mBuilding CNN Model','blue'))

Building CNN Model

cnn_sgd = Sequential()

cnn_sgd.add(tf.keras.layers.InputLayer(input_shape=(64,64,3,)))

cnn_sgd.add(tf.keras.layers.Conv2D(64, kernel_size=(3,3), activation='relu'))

cnn_sgd.add(tf.keras.layers.MaxPool2D(pool_size = (2,2)))

cnn_sgd.add(tf.keras.layers.BatchNormalization())

cnn_sgd.add(tf.keras.layers.Conv2D(64, kernel_size=(3,3), strides = (1,1), activation='relu'))

cnn_sgd.add(tf.keras.layers.MaxPool2D(pool_size = (2,2)))

cnn_sgd.add(tf.keras.layers.BatchNormalization())

cnn_sgd.add(tf.keras.layers.Conv2D(128, kernel_size=(3,3), strides = (1,1), activation='relu'))

cnn_sgd.add(tf.keras.layers.MaxPool2D(pool_size = (2,2)))

cnn_sgd.add(tf.keras.layers.BatchNormalization())

cnn_sgd.add(tf.keras.layers.Conv2D(128, kernel_size=(3,3), strides = (1,1), activation='relu'))

cnn_sgd.add(tf.keras.layers.MaxPool2D(pool_size = (2,2)))

cnn_sgd.add(tf.keras.layers.GlobalMaxPool2D())

cnn_sgd.add(tf.keras.layers.Flatten())

cnn_sgd.add(tf.keras.layers.Dense(128, activation='relu'))

cnn_sgd.add(tf.keras.layers.Dropout(0.2))
cnn_sgd.add(tf.keras.layers.Dense(64, activation='relu'))

cnn_sgd.add(tf.keras.layers.BatchNormalization())

cnn_sgd.add(tf.keras.layers.Dense(12, activation='softmax'))

print(colored('\x1B[1mCompiling Model Using SGD Optimizer','blue'))
sgd = optimizers.SGD(lr=.001, decay=1e-6, momentum=0.9)
cnn_sgd.compile(loss='categorical_crossentropy', optimizer=sgd, metrics=['accuracy'])

Compiling Model Using SGD Optimizer

print(colored('\x1B[1mFitting Model Using SGD Optimizer','blue'))

Fitting Model Using SGD Optimizer

history1 = cnn_sgd.fit(augmented_train,epochs=50,shuffle=True,
                    steps_per_epoch = 70,
                    validation_data=augmented_val,
                    validation_steps = int(500/32))
score = cnn_sgd.evaluate(X_test_new, y_test_new, verbose=0)
score_train['CNN(SGD)'] = cnn_sgd.evaluate(X_train, y_train, verbose=0)
score_test['CNN(SGD) Test Score']=score
accuracy['CNN(SGD) Accuracy'] = score

Epoch 1/50
70/70 [==============================] - 31s 422ms/step - loss: 2.8754 - accuracy: 0.1056 - val_loss: 2.5048 - val_accuracy: 0.1729
Epoch 2/50
70/70 [==============================] - 29s 417ms/step - loss: 2.3584 - accuracy: 0.2099 - val_loss: 2.0380 - val_accuracy: 0.3146
Epoch 3/50
70/70 [==============================] - 29s 418ms/step - loss: 2.1266 - accuracy: 0.2812 - val_loss: 2.0462 - val_accuracy: 0.2875
Epoch 4/50
70/70 [==============================] - 29s 417ms/step - loss: 2.0397 - accuracy: 0.2819 - val_loss: 2.6113 - val_accuracy: 0.2208
Epoch 5/50
70/70 [==============================] - 30s 423ms/step - loss: 1.9194 - accuracy: 0.3245 - val_loss: 1.8213 - val_accuracy: 0.3917
Epoch 6/50
70/70 [==============================] - 29s 417ms/step - loss: 1.8275 - accuracy: 0.3750 - val_loss: 2.5301 - val_accuracy: 0.2521
Epoch 7/50
70/70 [==============================] - 29s 418ms/step - loss: 1.7961 - accuracy: 0.3921 - val_loss: 2.2580 - val_accuracy: 0.2917
Epoch 8/50
70/70 [==============================] - 29s 418ms/step - loss: 1.7167 - accuracy: 0.4277 - val_loss: 2.7009 - val_accuracy: 0.2604
Epoch 9/50
70/70 [==============================] - 29s 420ms/step - loss: 1.6953 - accuracy: 0.4010 - val_loss: 1.6451 - val_accuracy: 0.4250
Epoch 10/50
70/70 [==============================] - 29s 417ms/step - loss: 1.6340 - accuracy: 0.4516 - val_loss: 1.5790 - val_accuracy: 0.4396
Epoch 11/50
70/70 [==============================] - 30s 422ms/step - loss: 1.5688 - accuracy: 0.4704 - val_loss: 1.6604 - val_accuracy: 0.4250
Epoch 12/50
70/70 [==============================] - 29s 418ms/step - loss: 1.5601 - accuracy: 0.4859 - val_loss: 1.6319 - val_accuracy: 0.4396
Epoch 13/50
70/70 [==============================] - 30s 425ms/step - loss: 1.4740 - accuracy: 0.5015 - val_loss: 1.3564 - val_accuracy: 0.5188
Epoch 14/50
70/70 [==============================] - 29s 420ms/step - loss: 1.4346 - accuracy: 0.5181 - val_loss: 1.5714 - val_accuracy: 0.4833
Epoch 15/50
70/70 [==============================] - 30s 430ms/step - loss: 1.3894 - accuracy: 0.5258 - val_loss: 1.7420 - val_accuracy: 0.4167
Epoch 16/50
70/70 [==============================] - 30s 434ms/step - loss: 1.3804 - accuracy: 0.5355 - val_loss: 1.4279 - val_accuracy: 0.5063
Epoch 17/50
70/70 [==============================] - 30s 427ms/step - loss: 1.3193 - accuracy: 0.5629 - val_loss: 1.5320 - val_accuracy: 0.4854
Epoch 18/50
70/70 [==============================] - 29s 418ms/step - loss: 1.2830 - accuracy: 0.5600 - val_loss: 1.4637 - val_accuracy: 0.5312
Epoch 19/50
70/70 [==============================] - 29s 420ms/step - loss: 1.2869 - accuracy: 0.5555 - val_loss: 1.4007 - val_accuracy: 0.5208
Epoch 20/50
70/70 [==============================] - 30s 422ms/step - loss: 1.2391 - accuracy: 0.5865 - val_loss: 1.5078 - val_accuracy: 0.4875
Epoch 21/50
70/70 [==============================] - 30s 422ms/step - loss: 1.2550 - accuracy: 0.5724 - val_loss: 1.1548 - val_accuracy: 0.5854
Epoch 22/50
70/70 [==============================] - 30s 426ms/step - loss: 1.2713 - accuracy: 0.5651 - val_loss: 1.3348 - val_accuracy: 0.5229
Epoch 23/50
70/70 [==============================] - 30s 426ms/step - loss: 1.2211 - accuracy: 0.6027 - val_loss: 1.2448 - val_accuracy: 0.5479
Epoch 24/50
70/70 [==============================] - 29s 418ms/step - loss: 1.1791 - accuracy: 0.5872 - val_loss: 1.0264 - val_accuracy: 0.6333
Epoch 25/50
70/70 [==============================] - 30s 426ms/step - loss: 1.1611 - accuracy: 0.5967 - val_loss: 1.0965 - val_accuracy: 0.6271
Epoch 26/50
70/70 [==============================] - 30s 426ms/step - loss: 1.1690 - accuracy: 0.6131 - val_loss: 1.3625 - val_accuracy: 0.5500
Epoch 27/50
70/70 [==============================] - 29s 420ms/step - loss: 1.1192 - accuracy: 0.6194 - val_loss: 1.1505 - val_accuracy: 0.5875
Epoch 28/50
70/70 [==============================] - 30s 424ms/step - loss: 1.1371 - accuracy: 0.6255 - val_loss: 1.0191 - val_accuracy: 0.6687
Epoch 29/50
70/70 [==============================] - 29s 421ms/step - loss: 1.0689 - accuracy: 0.6348 - val_loss: 1.2834 - val_accuracy: 0.5688
Epoch 30/50
70/70 [==============================] - 30s 426ms/step - loss: 1.0391 - accuracy: 0.6542 - val_loss: 1.2473 - val_accuracy: 0.5813
Epoch 31/50
70/70 [==============================] - 29s 420ms/step - loss: 1.0481 - accuracy: 0.6446 - val_loss: 1.6707 - val_accuracy: 0.4521
Epoch 32/50
70/70 [==============================] - 30s 425ms/step - loss: 1.1012 - accuracy: 0.6332 - val_loss: 1.0661 - val_accuracy: 0.6292
Epoch 33/50
70/70 [==============================] - 30s 426ms/step - loss: 1.0391 - accuracy: 0.6496 - val_loss: 0.9905 - val_accuracy: 0.6500
Epoch 34/50
70/70 [==============================] - 30s 424ms/step - loss: 0.9903 - accuracy: 0.6736 - val_loss: 1.0012 - val_accuracy: 0.6521
Epoch 35/50
70/70 [==============================] - 29s 421ms/step - loss: 1.0920 - accuracy: 0.6328 - val_loss: 1.1253 - val_accuracy: 0.6229
Epoch 36/50
70/70 [==============================] - 29s 419ms/step - loss: 1.0260 - accuracy: 0.6539 - val_loss: 1.2365 - val_accuracy: 0.5896
Epoch 37/50
70/70 [==============================] - 30s 428ms/step - loss: 0.9643 - accuracy: 0.6764 - val_loss: 1.0101 - val_accuracy: 0.6896
Epoch 38/50
70/70 [==============================] - 30s 421ms/step - loss: 0.9574 - accuracy: 0.6691 - val_loss: 1.2936 - val_accuracy: 0.5688
Epoch 39/50
70/70 [==============================] - 30s 426ms/step - loss: 0.9723 - accuracy: 0.6759 - val_loss: 1.1215 - val_accuracy: 0.5792
Epoch 40/50
70/70 [==============================] - 29s 419ms/step - loss: 0.9408 - accuracy: 0.6789 - val_loss: 0.8873 - val_accuracy: 0.6708
Epoch 41/50
70/70 [==============================] - 30s 425ms/step - loss: 0.8960 - accuracy: 0.7065 - val_loss: 0.9994 - val_accuracy: 0.6771
Epoch 42/50
70/70 [==============================] - 30s 424ms/step - loss: 0.9178 - accuracy: 0.6883 - val_loss: 0.9378 - val_accuracy: 0.6729
Epoch 43/50
70/70 [==============================] - 30s 425ms/step - loss: 0.8745 - accuracy: 0.7093 - val_loss: 0.9532 - val_accuracy: 0.6812
Epoch 44/50
70/70 [==============================] - 30s 422ms/step - loss: 0.8170 - accuracy: 0.7107 - val_loss: 0.9280 - val_accuracy: 0.6896
Epoch 45/50
70/70 [==============================] - 30s 429ms/step - loss: 0.8648 - accuracy: 0.7098 - val_loss: 1.1847 - val_accuracy: 0.6208
Epoch 46/50
70/70 [==============================] - 29s 421ms/step - loss: 0.8760 - accuracy: 0.6941 - val_loss: 1.1757 - val_accuracy: 0.5875
Epoch 47/50
70/70 [==============================] - 29s 421ms/step - loss: 0.8240 - accuracy: 0.7233 - val_loss: 1.0242 - val_accuracy: 0.6458
Epoch 48/50
70/70 [==============================] - 30s 425ms/step - loss: 0.9330 - accuracy: 0.6804 - val_loss: 0.8540 - val_accuracy: 0.6979
Epoch 49/50
70/70 [==============================] - 30s 422ms/step - loss: 0.8707 - accuracy: 0.7126 - val_loss: 0.8491 - val_accuracy: 0.7083
Epoch 50/50
70/70 [==============================] - 29s 420ms/step - loss: 0.8178 - accuracy: 0.7110 - val_loss: 0.9085 - val_accuracy: 0.6833

print(colored('\x1B[1mScore','green'),score)

Score [0.6512691378593445, 0.7853403091430664]

print(colored('\x1B[1m\t     Loss analysis graph of Model','green'))
plt.plot(history1.history['loss'])
plt.plot(history1.history['val_loss'])
plt.title('Model Loss')
plt.ylabel('Loss')
plt.xlabel('Epochs')
plt.legend(['train', 'test'])
plt.show()

	     Loss analysis graph of Model

print(colored('\x1B[1m\t  Accuracy analysis graph of Model','green'))
plt.plot(np.array(history1.history['accuracy']) * 100)
plt.plot(np.array(history1.history['val_accuracy']) * 100)
plt.ylabel('accuracy')
plt.xlabel('epochs')
plt.legend(['train', 'validation'])
plt.title('Accuracy over epochs')
plt.show()

	  Accuracy analysis graph of Model

print(colored('\x1B[1mEvaluating model against test sets','blue'))
scores = cnn_sgd.evaluate(X_test_new, y_test_new)
print('Test loss:', scores[0])
print('Test accuracy:', scores[1])

Evaluating model against test sets
6/6 [==============================] - 1s 91ms/step - loss: 0.6513 - accuracy: 0.7853
Test loss: 0.6512691378593445
Test accuracy: 0.7853403091430664

def plot_confusion_matrix(cm, classes, normalize=False,title='Confusion matrix',cmap=plt.cm.Greens):
    
    fig = plt.figure(figsize=(10,10))
    plt.imshow(cm, interpolation='nearest', cmap=cmap)
    plt.title(title)
    tick_marks = np.arange(len(classes))
    plt.xticks(tick_marks, classes, rotation=90)
    plt.yticks(tick_marks, classes)

    if normalize:
        cm = cm.astype('float') / cm.sum(axis=1)[:, np.newaxis]

    thresh = cm.max() / 2.
    for i, j in itertools.product(range(cm.shape[0]), range(cm.shape[1])):
        plt.text(j, i, cm[i, j],
                 horizontalalignment="center",
                 color="white" if cm[i, j] > thresh else "black")

    plt.tight_layout()
    plt.ylabel('Actual label')
    plt.xlabel('Predicted label')

print(colored('\x1B[1mPredict the values from the test data','blue'))
y_pred = cnn_sgd.predict(X_test_new)
y_pred_Classes = np.argmax(y_pred, axis = 1) 
trueY = np.argmax(y_test_new, axis = 1) 

Predict the values from the test data

print(colored('\x1B[1mConfusion Matrix of Model','blue'))
confusionMTX = confusion_matrix(trueY, y_pred_Classes) 
plot_confusion_matrix(confusionMTX, classes = labels.classes_) 

Confusion Matrix of Model

print(colored('\x1B[1mFinal score and accuracy of the model','blue'))
score, acc = cnn_sgd.evaluate(X_test_new,y_test_new)
score1, acc1 = cnn_sgd.evaluate(X_train,y_train)
print('Test score:', score,'   Test accuracy:', acc)
print('Train score:', score1,'   Train accuracy:',acc1)

Final score and accuracy of the model
6/6 [==============================] - 1s 93ms/step - loss: 0.6513 - accuracy: 0.7853
120/120 [==============================] - 11s 89ms/step - loss: 0.6428 - accuracy: 0.7687
Test score: 0.6512691378593445    Test accuracy: 0.7853403091430664
Train score: 0.6428350210189819    Train accuracy: 0.768686056137085

print(colored('\x1B[1mPrecicited Label','blue'))
pred_str

Precicited Label

array(['Fat Hen'], dtype=object)

cnn_adam = Sequential()

cnn_adam.add(tf.keras.layers.InputLayer(input_shape=(64,64,3,)))

cnn_adam.add(tf.keras.layers.Conv2D(64, kernel_size=(3,3), activation='relu'))

cnn_adam.add(tf.keras.layers.MaxPool2D(pool_size = (2,2)))

cnn_adam.add(tf.keras.layers.BatchNormalization())

cnn_adam.add(tf.keras.layers.Conv2D(64, kernel_size=(3,3), strides = (1,1), activation='relu'))

cnn_adam.add(tf.keras.layers.MaxPool2D(pool_size = (2,2)))

cnn_adam.add(tf.keras.layers.BatchNormalization())

cnn_adam.add(tf.keras.layers.Conv2D(128, kernel_size=(3,3), strides = (1,1), activation='relu'))

cnn_adam.add(tf.keras.layers.MaxPool2D(pool_size = (2,2)))

cnn_adam.add(tf.keras.layers.BatchNormalization())

cnn_adam.add(tf.keras.layers.Conv2D(128, kernel_size=(3,3), strides = (1,1), activation='relu'))

cnn_adam.add(tf.keras.layers.MaxPool2D(pool_size = (2,2)))

cnn_adam.add(tf.keras.layers.GlobalMaxPool2D())

cnn_adam.add(tf.keras.layers.Flatten())

cnn_adam.add(tf.keras.layers.Dense(128, activation='relu'))

cnn_adam.add(tf.keras.layers.Dropout(0.2))
cnn_adam.add(tf.keras.layers.Dense(64, activation='relu'))

cnn_adam.add(tf.keras.layers.BatchNormalization())

cnn_adam.add(tf.keras.layers.Dense(12, activation='softmax'))

print(colored('\x1B[1mCompiling Model Using ADAM Optimizer','blue'))
opt = Adam(lr=0.001, beta_1=0.9, beta_2=0.999, epsilon=None, decay=0.001, amsgrad=False)
cnn_adam.compile(optimizer = opt, loss = 'categorical_crossentropy', metrics = ['accuracy'])

Compiling Model Using ADAM Optimizer

print(colored('\x1B[1mFitting Model Using ADAM Optimizer','blue'))

Fitting Model Using ADAM Optimizer

history2 = cnn_adam.fit(augmented_train,epochs=50,shuffle=True,
                    steps_per_epoch = 40,
                    validation_data=augmented_val,
                    validation_steps = int(500/32))
score = cnn_adam.evaluate(X_test_new, y_test_new, verbose=0)
score_train['CNN(ADAM)'] = cnn_adam.evaluate(X_train, y_train, verbose=0)
score_test['CNN(ADAM) Test Score']=score
accuracy['CNN(ADAM) Accuracy'] = score

Epoch 1/50
40/40 [==============================] - 24s 573ms/step - loss: 2.7966 - accuracy: 0.1281 - val_loss: 3.3063 - val_accuracy: 0.1500
Epoch 2/50
40/40 [==============================] - 19s 463ms/step - loss: 2.2296 - accuracy: 0.2687 - val_loss: 2.1476 - val_accuracy: 0.3042
Epoch 3/50
40/40 [==============================] - 19s 464ms/step - loss: 1.8960 - accuracy: 0.3304 - val_loss: 1.8344 - val_accuracy: 0.3625
Epoch 4/50
40/40 [==============================] - 18s 460ms/step - loss: 1.7493 - accuracy: 0.4072 - val_loss: 2.2481 - val_accuracy: 0.3604
Epoch 5/50
40/40 [==============================] - 18s 458ms/step - loss: 1.6274 - accuracy: 0.4532 - val_loss: 1.4938 - val_accuracy: 0.4938
Epoch 6/50
40/40 [==============================] - 18s 452ms/step - loss: 1.5883 - accuracy: 0.4610 - val_loss: 1.7393 - val_accuracy: 0.4500
Epoch 7/50
40/40 [==============================] - 18s 453ms/step - loss: 1.5563 - accuracy: 0.4710 - val_loss: 1.8622 - val_accuracy: 0.3979
Epoch 8/50
40/40 [==============================] - 18s 453ms/step - loss: 1.4994 - accuracy: 0.4840 - val_loss: 1.6910 - val_accuracy: 0.3917
Epoch 9/50
40/40 [==============================] - 18s 458ms/step - loss: 1.2695 - accuracy: 0.5787 - val_loss: 1.4376 - val_accuracy: 0.5229
Epoch 10/50
40/40 [==============================] - 18s 448ms/step - loss: 1.2793 - accuracy: 0.5640 - val_loss: 1.2595 - val_accuracy: 0.5583
Epoch 11/50
40/40 [==============================] - 18s 445ms/step - loss: 1.2844 - accuracy: 0.5734 - val_loss: 1.4653 - val_accuracy: 0.4875
Epoch 12/50
40/40 [==============================] - 18s 448ms/step - loss: 1.2302 - accuracy: 0.5844 - val_loss: 1.4722 - val_accuracy: 0.5021
Epoch 13/50
40/40 [==============================] - 18s 455ms/step - loss: 1.1122 - accuracy: 0.6268 - val_loss: 1.4525 - val_accuracy: 0.5167
Epoch 14/50
40/40 [==============================] - 18s 446ms/step - loss: 1.1516 - accuracy: 0.6050 - val_loss: 1.3902 - val_accuracy: 0.5229
Epoch 15/50
40/40 [==============================] - 18s 459ms/step - loss: 1.1218 - accuracy: 0.6193 - val_loss: 1.1067 - val_accuracy: 0.6333
Epoch 16/50
40/40 [==============================] - 18s 463ms/step - loss: 1.0712 - accuracy: 0.6490 - val_loss: 1.3630 - val_accuracy: 0.5271
Epoch 17/50
40/40 [==============================] - 18s 459ms/step - loss: 0.9883 - accuracy: 0.6743 - val_loss: 0.9133 - val_accuracy: 0.6979
Epoch 18/50
40/40 [==============================] - 19s 466ms/step - loss: 0.9372 - accuracy: 0.6518 - val_loss: 1.0688 - val_accuracy: 0.6438
Epoch 19/50
40/40 [==============================] - 18s 453ms/step - loss: 0.9898 - accuracy: 0.6891 - val_loss: 0.9479 - val_accuracy: 0.6646
Epoch 20/50
40/40 [==============================] - 19s 464ms/step - loss: 1.0246 - accuracy: 0.6487 - val_loss: 0.9297 - val_accuracy: 0.6833
Epoch 21/50
40/40 [==============================] - 18s 459ms/step - loss: 0.8762 - accuracy: 0.6955 - val_loss: 1.2643 - val_accuracy: 0.5625
Epoch 22/50
40/40 [==============================] - 18s 462ms/step - loss: 0.9842 - accuracy: 0.6513 - val_loss: 0.8806 - val_accuracy: 0.6938
Epoch 23/50
40/40 [==============================] - 18s 462ms/step - loss: 0.9120 - accuracy: 0.6894 - val_loss: 0.8272 - val_accuracy: 0.7271
Epoch 24/50
40/40 [==============================] - 20s 496ms/step - loss: 0.9117 - accuracy: 0.6771 - val_loss: 1.1133 - val_accuracy: 0.6083
Epoch 25/50
40/40 [==============================] - 19s 464ms/step - loss: 0.8073 - accuracy: 0.7336 - val_loss: 0.9893 - val_accuracy: 0.6583
Epoch 26/50
40/40 [==============================] - 19s 465ms/step - loss: 0.8546 - accuracy: 0.7078 - val_loss: 1.0728 - val_accuracy: 0.6125
Epoch 27/50
40/40 [==============================] - 18s 462ms/step - loss: 0.8446 - accuracy: 0.7138 - val_loss: 0.7463 - val_accuracy: 0.7396
Epoch 28/50
40/40 [==============================] - 19s 465ms/step - loss: 0.7573 - accuracy: 0.7295 - val_loss: 0.8089 - val_accuracy: 0.7063
Epoch 29/50
40/40 [==============================] - 19s 467ms/step - loss: 0.8174 - accuracy: 0.7274 - val_loss: 1.0660 - val_accuracy: 0.6313
Epoch 30/50
40/40 [==============================] - 18s 461ms/step - loss: 0.7840 - accuracy: 0.7357 - val_loss: 0.7414 - val_accuracy: 0.7292
Epoch 31/50
40/40 [==============================] - 18s 449ms/step - loss: 0.7384 - accuracy: 0.7442 - val_loss: 1.0038 - val_accuracy: 0.6500
Epoch 32/50
40/40 [==============================] - 19s 465ms/step - loss: 0.7251 - accuracy: 0.7620 - val_loss: 0.8133 - val_accuracy: 0.7083
Epoch 33/50
40/40 [==============================] - 19s 470ms/step - loss: 0.7052 - accuracy: 0.7533 - val_loss: 0.8014 - val_accuracy: 0.7542
Epoch 34/50
40/40 [==============================] - 18s 461ms/step - loss: 0.7120 - accuracy: 0.7796 - val_loss: 0.9761 - val_accuracy: 0.6938
Epoch 35/50
40/40 [==============================] - 19s 480ms/step - loss: 0.7021 - accuracy: 0.7609 - val_loss: 0.8480 - val_accuracy: 0.6917
Epoch 36/50
40/40 [==============================] - 18s 457ms/step - loss: 0.6562 - accuracy: 0.7702 - val_loss: 0.6363 - val_accuracy: 0.7792
Epoch 37/50
40/40 [==============================] - 18s 452ms/step - loss: 0.6583 - accuracy: 0.7827 - val_loss: 0.9758 - val_accuracy: 0.6625
Epoch 38/50
40/40 [==============================] - 19s 464ms/step - loss: 0.7659 - accuracy: 0.7389 - val_loss: 1.1262 - val_accuracy: 0.6083
Epoch 39/50
40/40 [==============================] - 18s 460ms/step - loss: 0.7147 - accuracy: 0.7530 - val_loss: 0.6087 - val_accuracy: 0.7937
Epoch 40/50
40/40 [==============================] - 18s 456ms/step - loss: 0.7505 - accuracy: 0.7507 - val_loss: 0.8719 - val_accuracy: 0.6917
Epoch 41/50
40/40 [==============================] - 18s 447ms/step - loss: 0.7648 - accuracy: 0.7203 - val_loss: 0.6236 - val_accuracy: 0.7875
Epoch 42/50
40/40 [==============================] - 18s 448ms/step - loss: 0.6642 - accuracy: 0.7653 - val_loss: 0.8757 - val_accuracy: 0.6979
Epoch 43/50
40/40 [==============================] - 18s 445ms/step - loss: 0.6301 - accuracy: 0.7844 - val_loss: 0.6644 - val_accuracy: 0.7625
Epoch 44/50
40/40 [==============================] - 18s 446ms/step - loss: 0.7278 - accuracy: 0.7587 - val_loss: 0.8367 - val_accuracy: 0.7083
Epoch 45/50
40/40 [==============================] - 18s 457ms/step - loss: 0.6007 - accuracy: 0.7971 - val_loss: 0.6185 - val_accuracy: 0.7708
Epoch 46/50
40/40 [==============================] - 18s 454ms/step - loss: 0.5511 - accuracy: 0.8067 - val_loss: 0.7673 - val_accuracy: 0.7354
Epoch 47/50
40/40 [==============================] - 18s 446ms/step - loss: 0.6566 - accuracy: 0.7742 - val_loss: 0.6329 - val_accuracy: 0.7812
Epoch 48/50
40/40 [==============================] - 18s 456ms/step - loss: 0.5850 - accuracy: 0.7996 - val_loss: 0.7346 - val_accuracy: 0.7458
Epoch 49/50
40/40 [==============================] - 18s 456ms/step - loss: 0.5862 - accuracy: 0.7985 - val_loss: 0.5892 - val_accuracy: 0.7937
Epoch 50/50
40/40 [==============================] - 18s 452ms/step - loss: 0.6033 - accuracy: 0.7966 - val_loss: 0.4527 - val_accuracy: 0.8354

print(colored('\x1B[1mScore','green'),score)

Score [0.3903430998325348, 0.8638743162155151]

print(colored('\x1B[1m\t     Loss analysis graph of Model','green'))
plt.plot(history2.history['loss'])
plt.plot(history2.history['val_loss'])
plt.title('Model Loss')
plt.ylabel('Loss')
plt.xlabel('Epochs')
plt.legend(['train', 'test'])
plt.show()

	     Loss analysis graph of Model

print(colored('\x1B[1m\t  Accuracy analysis graph of Model','green'))
plt.plot(np.array(history2.history['accuracy']) * 100)
plt.plot(np.array(history2.history['val_accuracy']) * 100)
plt.ylabel('accuracy')
plt.xlabel('epochs')
plt.legend(['train', 'validation'])
plt.title('Accuracy over epochs')
plt.show()

	  Accuracy analysis graph of Model

print(colored('\x1B[1mEvaluating model against test sets','blue'))
scores = cnn_adam.evaluate(X_test_new, y_test_new)
print('Test loss:', scores[0])
print('Test accuracy:', scores[1])

Evaluating model against test sets
6/6 [==============================] - 1s 90ms/step - loss: 0.3903 - accuracy: 0.8639
Test loss: 0.3903430998325348
Test accuracy: 0.8638743162155151

def plot_confusion_matrix(cm, classes, normalize=False,title='Confusion matrix',cmap=plt.cm.Greens):
    
    fig = plt.figure(figsize=(10,10))
    plt.imshow(cm, interpolation='nearest', cmap=cmap)
    plt.title(title)
    tick_marks = np.arange(len(classes))
    plt.xticks(tick_marks, classes, rotation=90)
    plt.yticks(tick_marks, classes)

    if normalize:
        cm = cm.astype('float') / cm.sum(axis=1)[:, np.newaxis]

    thresh = cm.max() / 2.
    for i, j in itertools.product(range(cm.shape[0]), range(cm.shape[1])):
        plt.text(j, i, cm[i, j],
                 horizontalalignment="center",
                 color="white" if cm[i, j] > thresh else "black")

    plt.tight_layout()
    plt.ylabel('Actual label')
    plt.xlabel('Predicted label')

print(colored('\x1B[1mPredict the values from the test data','blue'))
y_pred = cnn_adam.predict(X_test_new)
y_pred_Classes = np.argmax(y_pred, axis = 1) 
trueY = np.argmax(y_test_new, axis = 1) 

Predict the values from the test data

print(colored('\x1B[1mConfusion Matrix of Model','blue'))
confusionMTX = confusion_matrix(trueY, y_pred_Classes) 
plot_confusion_matrix(confusionMTX, classes = labels.classes_) 

Confusion Matrix of Model

print(colored('\x1B[1mFinal score and accuracy of the model','blue'))
score, acc = cnn_adam.evaluate(X_test_new,y_test_new)
score1, acc1 = cnn_adam.evaluate(X_train,y_train)
print('Test score:', score,'   Test accuracy:', acc)
print('Train score:', score1,'   Train accuracy:',acc1)
#Model with training accuracy of 86% while testing accuracy 84%.

Final score and accuracy of the model
6/6 [==============================] - 1s 94ms/step - loss: 0.3903 - accuracy: 0.8639
120/120 [==============================] - 11s 91ms/step - loss: 0.4405 - accuracy: 0.8403
Test score: 0.3903430998325348    Test accuracy: 0.8638743162155151
Train score: 0.4404798448085785    Train accuracy: 0.840283215045929

print(colored('\x1B[1mPredection using test folder images','blue'))
test_images_path= '/content/drive/MyDrive/Files/Part !/Part-1 - Plant Seedling Classification Data/Seedling - Prediction/*.png'
test_images = glob(test_images_path)
test_images_arr = []
print(colored('\x1B[1mGetting the images from test folder','green'))
for img in test_images:
    test_images_arr.append(cv2.resize(cv2.imread(img), (64, 64)))
test_X = np.asarray(test_images_arr)
print(colored('\x1B[1mNormalizing the Images','green'))
test_X = test_X.astype('float32') / 255

Predection using test folder images
Getting the images from test folder
Normalizing the Images

print('''\n\033[1m''' + '''Displaying and comparing all the models designed with their train and test accuracies''' + '''\033[0m''')
results = pd.DataFrame({'Model' : score_train.keys(),
                        'Train Score' : score_train.values(),
                        'Test Score': score_test.values(),
                        'Accuracy' : accuracy.values()})
results

Displaying and comparing all the models designed with their train and test accuracies

	Model	Train Score	Test Score	Accuracy
0	SVM	0.996103	0.499651	0.499651
1	Random Forest	0.958933	0.427673	0.427673
2	Decision Tree	0.264388	0.22362	0.22362
3	NN(SGD)	[0.9962068796157837, 0.6698138117790222]	[1.4598736763000488, 0.5073375105857849]	[1.4598736763000488, 0.5073375105857849]
4	NN(ADAM)	[0.35462790727615356, 0.9045370817184448]	[1.6843140125274658, 0.5094339847564697]	[1.6843140125274658, 0.5094339847564697]
5	CNN(SGD)	[0.6428350210189819, 0.768686056137085]	[0.6512691378593445, 0.7853403091430664]	[0.6512691378593445, 0.7853403091430664]
6	CNN(ADAM)	[0.4404798448085785, 0.840283215045929]	[0.3903430998325348, 0.8638743162155151]	[0.3903430998325348, 0.8638743162155151]

From the table it is clear that CNN model using ADAM as optimizer gives the best accuracy of all models

But CNN model using SGD optimizer also gives considerable accuracy

print(colored('\x1B[1mSaving Model:','blue'))
cnn_adam.save('CNN_ADAM Classification') 

Saving Model:
INFO:tensorflow:Assets written to: CNN_ADAM Classification/assets

print(colored('\x1B[1mPredection using test folder images','blue'))
test_images_path= '/content/drive/MyDrive/Files/Part !/Part-1 - Plant Seedling Classification Data/Seedling - Prediction/*.png'
test_images = glob(test_images_path)
test_images_arr = []
print(colored('\x1B[1mGetting the images from test folder','green'))
for img in test_images:
    test_images_arr.append(cv2.resize(cv2.imread(img), (64, 64)))
test_X = np.asarray(test_images_arr)
print(colored('\x1B[1mNormalizing the Images','green'))
test_X = test_X.astype('float32') / 255

Predection using test folder images
Getting the images from test folder
Normalizing the Images

print(colored('\x1B[1mPredicting using CNN_ADAM model','blue'))
predictions = cnn_adam.predict(test_X)
preds = np.argmax(predictions, axis=1)
pred_str = labels.classes_[preds]
pred_str

Predicting using CNN_ADAM model

array(['Loose Silky-bent'], dtype=object)

print(colored('\x1B[1mPredicting using CNN_SGD model','blue'))
predictions = cnn_sgd.predict(test_X)
preds = np.argmax(predictions, axis=1)
pred_str = labels.classes_[preds]
pred_str

Predicting using CNN_SGD model

array(['Fat Hen'], dtype=object)

Explain in depth why CNN out performs neural networks which in turn out perform supervised learning models when it comes to image classification. Use the markdown option in Jupiter for your answer.

It is easy to use as it does not need much image processing unlike than Artificial Neural Network

The CNN Model just need to convert image format to array and normalize the pixel values

The Convolution Layer is a kernel which is spatially smaller than an image but has more details in-depth.

The Pooling Layer helps in reducing the spatial size of the representation(output), which decreases the required amount of computation and weights.

Fully Connected Layer as usual have full connectivity with all neurons in the preceding and succeeding layers